DescriptionRecent advancements in language models, such as ChatGPT, have brought excitement and concerns regarding the implications of language models on student learning and assessment in Higher Education. Proponents argue that recent advancements will enhance the student experience and improve academic achievement in higher education. Those concerned argue that language models will impede student learning and assessment, and they call for a cautious approach to adopting these tools in student learning and assessment. Therefore, since the release of ChatGPT, several tools have been developed to help the concerned parties detect artificially generated text, and these attempts have yet to be accurate and reliable. This review explores the literature on language models and student learning and assessment in higher education to encourage future research. A search protocol was employed using the Scopus database to identify 24 articles based on relevant keywords and article selection criteria. The analysis adopted 'Reflexive Thematic Analysis'. The reflexive themes are explored from the selected list of articles, and these themes suggest that language models may alter how students learn and how their learning is assessed. The themes also draw attention to the drawbacks of language models, such as the risk of bias and inaccurate information and the need for students to be aware of the quality of the information these models generate. Propositions are included after each theme to encourage future research.
|28 Mar 2023
|Huddersfield Business School’s 8th Annual Research Conference
|Huddersfield, United Kingdom
|Degree of Recognition